The capability
See what matters.
Understand why.
Follow the evidence from the process to its subprocesses. Explore which explanation fits—and how the production outlook changes under alternative assumptions.
Synthetic evidence · 08:30
Inspection receives fewer engines
The downstream change is consistent with the upstream constraint.
01 / Explanation
Where does the evidence point?
Probability of the dominant explanation · brightness shows 0–100%
Subprocess shares form their parent. Displayed values are rounded; do not add parent and child rows together.
02 / Salience
Influence on the target outlook
Salience = the size of the prediction change when the selected explanation is excluded.
03 / Alternative predictions
What if this condition is absent?
Chance of meeting the same 14:00 target
Select a process or subprocess in the swim lanes.
Combined alternatives reweight the remaining explanations.
Excluding an explanation tests a model assumption. It does not show what repairing a fault would achieve.
Connect the observed pattern to the target and the next investigation.
Illustrative concept using synthetic events and authored model assumptions. Probabilities concern one dominant explanation; real faults can coexist. These are not validated product outputs or confirmed causes. The alternatives shown cover this demonstration’s scenario set.
From evidence to attention
A changing picture.
A clearer investigation.
Nentrope is developing an advisory layer to connect operational evidence, possible explanations and their consequences for a production target.
01
Follow the evidence.
Timestamped events, counts, transfers and error messages update the picture. Supporting and conflicting evidence can change which explanation fits.
02
Locate the explanation.
Explore probability across a process and its subprocesses. Keep competing explanations and uncertainty visible as the evidence develops.
03
Test what matters.
Compare target predictions with a selected explanation excluded. A large change identifies an assumption that deserves closer investigation.
Designed for your operation
Fits alongside
the existing stack.
MES, ERP, SCADA and historian data provide the starting evidence. The first step uses a historical export to assess whether the data and process are suitable.
The approach is advisory. Your team retains responsibility for operational decisions and machine control.
Prototype development and validation. The interaction above illustrates a capability direction; customer-specific warning performance and financial benefits remain to be demonstrated.
Questions
Before we begin.
Does an alarm prove the cause?
No. An error message is an observation. It can support several explanations, and later evidence can change their relative plausibility.
What does salience mean here?
In this illustration, salience is the absolute difference in target probability when a selected explanation is excluded and the remaining explanations are reweighted. It measures sensitivity to that assumption, not proof of cause or the benefit of a repair.
What data is useful?
Timestamped counts, events, transfers and relevant error messages, together with process structure and the production target. Available coverage and identifiers are reviewed before assuming suitability.
Do we need to replace our current systems?
The initial assessment uses a historical export from existing systems. Any later live integration is scoped separately. No changes to machine control are required for the assessment.
Start with one flow
One production flow.
One useful conversation.
A 30-minute conversation about one production flow where earlier visibility of risk could make a difference. No data or system access required.
Discuss one production flow